Integrated Circuits

As integrated circuits (ICs) become crowded with more semiconductor devices and internal connections, one wonders how they will be connected to the external world. We know also that cosmic rays and other particle radiation from the environment already disrupt some of the operations by random destruction, and these effects will become worse as the scale diminishes. However, neither connection to the external world via gold wires of any size nor the background particle radiation is the major problem today. What is?

Answer

Heat dissipation is the biggest problem in ICs. Simple, old-fashioned thermal energy limits the density of electronic components. The ability to miniaturize continues to improve, but unless thermal energy production per volume is decreased or new geometrical paths for thermal energy transport away from the sources are devised, the game is lost. At present, 3-D ICs offer a temporary reprieve, but even they will find their limit.

Some short time scale solutions may be possible. The best thermal conductor among crystalline materials is diamond, so going to a diamond substrate may be a solution. However, the technology of diamond is not yet competitive with silicon technology. Also, components on these substrates that require significantly less energy to function as a gate would delay the overwhelming impact of thermal problems. Optical information transfer between components would eliminate electrical currents and their thermal effects, but silicon does not have the right optical properties; hence the active research into doping silicon to make the desirable optical properties.

At the longer time scale of several decades, perhaps the silicon and semiconductor technology will simply fade away in favor of some other technology on the time horizon that seems unachievable today but that would become viable then. Or the newer technology hasn’t even been dreamed about yet!

For any solid or liquid material, quantum disturbances from cosmic rays may decide the ultimate limit in electronic component density unless redundancy can solve this problem. For optical systems based on light interference, and so on, who knows what is possible? Whatever wins in the decades ahead will be numerous orders of magnitude smaller and faster, as well as more robust than what we have today.